Dark_Pattern (Asian heart)

This repository contains code for detecting and analyzing dark patterns (NLP + Vision). Large datasets and model weights are excluded from the repository and should be stored externally or tracked with Git LFS.

Included (what to commit)

• All source code: app.py, Vision_/, NLP_/ (scripts, training and prediction code)
• requirements.txt and web_requirements.txt
• templates/ and static/ (web UI files)
• Notebooks for demonstration and experiments

Excluded (what NOT to commit)

• Virtual environment: dp_env/, venv/, .venv/
• Datasets: NLP_Dataset/, Vision_Dataset/
• Training outputs and model artifacts: model_out/, outputs/, runs/, *.pt, *.safetensors

Push instructions (example)

git init
git add .
git commit -m "Initial commit"
git branch -M main
git remote add origin https://github.com/baby-rex/Dark_Pattern.git
git push -u origin main

If you have large model files you want to keep in the repo, use Git LFS:

brew install git-lfs   # macOS
git lfs install
git lfs track "*.pt"
git lfs track "*.safetensors"
git add .gitattributes
git add .
git commit -m "Track large model files with Git LFS"
git push origin main

Architecture

High-level layout

• Root files

  • app.py — Flask/FastAPI (web) entrypoint for serving UI and APIs.
  • web_requirements.txt — dependencies for the web service.
  • yolov8m.pt — (large) object-detection weight (keep out of repo or use Git LFS).

• Web UI

  • templates/ — HTML templates for server-rendered pages.
  • static/ — CSS, JS, and static assets (style.css).

• NLP module (NLP_/)

  • train_model.py — training script for NLP models.
  • predict.py — inference/prediction utilities.
  • requirements.txt — module-specific dependencies.
  • misc.ipynb — exploratory notebook.

• Vision module (Vision_/)

  • train_model.py — training script for vision models.
  • prediction.py — inference utilities for images.
  • xai_test_out/ — explainability outputs and visualizations (generated).

• Datasets (excluded from repo)

  • NLP_Dataset/ — raw/processed NLP datasets (large; keep externally).
  • Vision_Dataset/ — image dataset, labels, and splits (large).

• Model artifacts & checkpoints (excluded)

  • model_out/ — saved model weights, checkpoints, scheduler/optimizer states.
  • runs/, outputs/ — training logs, tensorboard data, evaluation outputs.

• Environment & tools

  • dp_env/ — local Python virtual environment (do not commit).
  • requirements.txt, web_requirements.txt — top-level dependency lists.

Typical workflows

• Development: edit source under NLP_/ and Vision_/, run unit or integration checks locally using the dp_env virtualenv.
• Training: run train_model.py in the appropriate module; outputs and checkpoints write to model_out/ and runs/ (ignored).
• Inference / Serving: app.py loads model artifacts (prefer loading from external storage) and serves predictions using templates/ + static/.

Recommended files to commit

• All source code and small helper scripts: app.py, Vision_/, NLP_/ (code only — not large artifacts).
• templates/ and static/ (web UI assets).
• requirements.txt and web_requirements.txt.
• README.md and minimal manifest.txt (optional) listing external datasets and model locations.

Files to exclude from Git (already in .gitignore)

• dp_env/, NLP_Dataset/, Vision_Dataset/, model_out/, runs/, outputs/, large weights like *.pt/*.safetensors.

Large files strategy

• Use Git LFS for model weights you must track (*.pt, *.safetensors).
• Store datasets and heavy checkpoints in cloud storage (S3, Google Drive, or an artifact store) and provide download scripts.

Notes

• Before pushing, verify .gitignore excludes the large folders listed above.
• If you accidentally committed large files, use git rm --cached <file> and consider bfg or git filter-repo to remove them from history.